Multiple-day forecast of residential power consumption
Keywords:
Load forecasting, deep learning methods, RNN model, manual feature, forecasting horizon, recursive forecasting, multiple-day forecastingAbstract
Load forecasting is a crucial step in achieving optimal power exchange and ensuring the robust operation of utilities, particularly in the active integration of distributed renewable energy sources (DRES). Deep learning methods have gained popularity among researchers due to their capability to generalize complex nonlinearity in power consumption fluctuations. Especially, recurrent neural networks are suitable for modeling electricity consumption as they have strong temporal patterns. This work aims to continue enhancing RNN models, which are built on base and manual features. The forecasting horizon has been extended to meet the requirements of practical application. Two approaches of multiple-day forecasting have been compared, and the proposed method outperformed the alternative recursive forecasting method.
References
Tang, X., Dai, Y., Wang, T., & Chen, Y. Short-term power load forecasting based on multi-layer bidirectional recurrent neural network. IET Generation, Transmission & Distribution, 2019. IET Research Journal
Cai, C., Tao, Y., Zhu, T., & Deng, Z. Short-Term Load Forecasting Based on Deep Learning Bidirectional LSTM Neural Network. Applied Sciences, 2021. MDPI
Yu, K., et al. Adaptive Bi-Directional LSTM Short-Term Load Forecasting with Improved Attention Mechanisms. Energies, 2024. MDPI
Xiong, N., Li, M., Li, K., Xiao, Z., Liao, L., & Wang, L. (2023). Multi-feature Short-Term Power Load Prediction Method Based on Bidirectional LSTM Network. In: ICNC-FSKD 2022. Lecture Notes Springer.
Ke, K., Hongbin, S., Chengkang, Z., Sun, H., & Brown, C. Short-Term Electrical Load Forecasting Method Based on Stacked Auto-encoding and GRU Neural Network. Evolutionary Intelligence, 2019.
Bohara, B., Fernandez, R. I., Gollapudi, V., & Li, X. Short-Term Aggregated Residential Load Forecasting using BiLSTM and CNN-BiLSTM. arXiv preprint, 2023.
Dong, M., & Grumbach, L. S. Hybrid Distribution Feeder Long-Term Load Forecasting Method Based on Sequence Prediction. arXiv preprint, 2018.
In, Y., & Jung, J. Y. Simple averaging of direct and recursive forecasts via partial pooling using machine learning. International Journal of Forecasting, 2022.
24-Step Short-term Power Load Forecasting Model Utilizing KOA-BiTCN-BiGRU-Attentions[v1]. Preprints.org, 2024.
Hybrid feature-based neural network regression method for load profiles forecasting. Energy Informatics, 2025.
G. Chevillon, “Direct multi-step estimation and forecasting,” 2007. doi: 10.1111/j.14676419.2007.00518.x.
Abdurazakov, Nosirbek N., et al. “Feature Engineering for Short Term Residential Load Forecasting Using RNN-Based Neural Networks.” International Conference on Environmental Design, Material Science, and Engineering Technologies. Cham: Springer Nature Switzerland, 2024.
Keras documentation on Time Distributed Layer. https://keras.io/api/layers/recurrent_layers/time_distributed/
